Estimating the Soil Infiltration in Treated Wastewater Irrigated Land Using Artificial Intelligence

Rooshenas, N. (2013) Estimating the Soil Infiltration in Treated Wastewater Irrigated Land Using Artificial Intelligence. Masters thesis, University of Zabol.

Estimating the Soil Infiltration in.pdf

Download (210kB) | Preview


Water quality is one of the most effective factors on soil infiltration. The widespread use of wastewater in recent years caused the accurate estimation of soil infiltration to be essential for achieving proper irrigation management. Therefore, Philip, Kostiakov, Kostiakov-Lewis, Green-Ampt, Horton, SCS and linear regression models for soil infiltration estimation were evaluated. Infiltration experiments were carried out in an area of 190 ha of agricultural lands under wheat cultivation, around the Zabol refinery of municipal waste water for six soil texture classes including silty clay, clay loam, sandy loam, loam, loamy sand, silty clay loam. These lands are irrigated with treated wastewater for over 10 years. The study area was divided into pixels with 150m 150m size. The cumulative infiltration was determined using double ring experiment. Some physicochemical soil properties were determined in soil samples taken from the center of each pixel. The best performance in the estimation of infiltration have been 6-variable transfer functions in the region. Kostiakov and Horton models were the best profitable models for silty clay soils. Whereas, the others were the best for cumulative infiltration estimation in silty clay loam soils. Green-Ampt and Kostiakov models were the best fitted on the observed cumulative infiltration data with root mean square, mean bias error and efficient coefficient of 0.16 and 0.15 cm, -0.03 and 0.04 cm, and 0.74 and 0.78 cm, respectively. The considerable reduction in soil infiltration due to over accumulation of salt and sodium in the surface soil layers through long term application of waste water in the study area. Neural network model with 5 variables had the highest correlation coefficient (R=0.98) and The least root mean square error (RMSE=0.083) for estimating soil infiltration. Long-term and unmanaged use of treated wastewater aaused accumulation of excessive salt and sodium in the soil surface layer of soil and therefore the significant reduction infiltration for soil textures classes.

Item Type: Thesis (Masters)
Uncontrolled Keywords: transfer function, exchangeable sodium, neural network, wheat, wastewater, infiltration models
Subjects: S Agriculture > S Agriculture (General)
Depositing User: admin admin1 admin2
Date Deposited: 15 Jun 2016 07:57
Last Modified: 15 Jun 2016 07:57

Actions (login required)

View Item View Item